• 82% of UK data centre projects delayed due to fibre availability constraints
  • Rail-based fibre expansion seen as key to unlocking AI infrastructure growth

What happened

Neos Networks CEO Lee Myall has warned that the UK’s AI strategy is being constrained by fibre infrastructure rather than compute capacity. His comments come amid rising government support for AI and quantum investment announced by Rachel Reeves.

Myall said industry attention has focused too heavily on computing power. He argued that land, energy and fibre connectivity are equally critical for building high-performance data centres.

Neos Networks research shows that 82% of UK data centre operators have delayed expansion or site builds due to limited fibre availability. In addition, 95% of operators say access to high-capacity fibre directly influences investment location decisions.

To address this constraint, Neos is expanding its national footprint through rail corridors. Its Project Reach initiative uses railway infrastructure to deploy long-distance fibre routes across the UK. The network spans key corridors linking London with Manchester, Bristol, Newcastle and Scotland, with large sections currently under construction.

Myall also noted that AI workloads are changing infrastructure needs. He said latency is becoming less critical, while proximity to power and land is becoming more important for next-generation data centres.

Why it’s important

The comments highlight a structural imbalance in the UK’s AI growth strategy. While policy and capital investment are accelerating compute deployment, physical network capacity is emerging as a limiting factor.

Fibre connectivity now acts as a gatekeeper for data centre expansion. When most operators delay projects due to network access, investment timelines shift away from financial planning and become infrastructure-dependent. This risks slowing down AI rollout even when funding is available.

The shift also reflects a broader change in how AI infrastructure is designed. Large-scale AI systems are less dependent on low-latency urban clusters and more focused on access to power and land. This increases demand for long-distance, high-capacity optical networks connecting regional sites.

Neos’s rail-based deployment model signals a practical response to this challenge. By using existing transport corridors, it reduces build friction and accelerates fibre rollout across multiple regions. This approach also aligns with a wider industry shift towards targeted infrastructure investment rather than speculative overbuild from earlier broadband cycles.

Ultimately, the UK’s ability to compete in AI will depend on synchronising compute, power and fibre infrastructure. Without faster connectivity expansion, even strong investment pipelines may struggle to translate into real-world AI capacity at scale.

Also read: Nokia builds SkyFiber fibre network in rural Nevada

Also read: CityFibre launches 8.5Gbps to lift UK broadband standard